Optimisation of Rietveld – Cement
Optimisation of Rietveld Based Methods for the On-line Analysis of Portland Cement and Clinker
The emergence of interest in XRD as a tool for monitoring mineralogical abundances in industrial mineral processing plants has led to the development of a range of plant based instrumentation for the analysis of dry sample materials. These instruments can be divided into two general types, namely:-
Portland cement is a mineralogically complex material consisting of some 10 to 15 phases of interest varying in concentration from 60 wt% down to ~ 0.2 wt%. The physical and chemical characteristics of the component phases, including the detailed chemical composition, crystallite size and ratio of polymorphs, are dependent on the nature of the starting materials and the production conditions present in a specific plant (Taylor, 1990; Madsen and Scarlett, 2000). Therefore, a robust XRD analytical regime is required to minimize the possibility of erroneous results being produced. In addition, use in an industrial processing plant often means that expert analysts are not available to check that the method is working correctly; this requires that the method operate reliably and report the results to the plant control room without operator intervention. Careful optimization of the analysis conditions and verification of results is essential.
In order to produce the most reliable analysis under these conditions, especially for the minor phases, the steps needed to develop a robust Rietveld-based analytical method include:-
1. The ferrite and aluminate phases, C4AF and C3A, are present in cement clinker at about 10 to 15 wt% in total (The cement industry nomenclature has been used when referring to the phases present in cement and clinker.
C = CaO, S = SiO2, A = Al2O3, F = Fe2O3). Hence, their diffraction patterns are normally dominated by the more abundant silicate phases, C3S and C2S. Taylor (1990) describes a method where C3S and C2S can be extracted from the sample using a mixture of salicylic acid and methanol (SAM) leaving a residue (denoted Residue #2) in which C4AF and C3A are the major phases remaining. Collection and analysis of high quality, laboratory based XRD data from this residue allows refinement of parameters for the C4AF and C3A. In addition, this pattern can be used to identify whether (i) C3A is present as more than one polymorph - cubic and orthorhombic are the most commonly occurring forms, and (ii) C4AF exhibits a range of compositions with varying Fe/Al ratios – some cement plants produce C4AF with two or more distinct compositions which have different unit cell dimensions and are readily analysed as separate phases. During on-line analysis, the crystallite size and unit cell parameters are fixed at the values determined from the residues or, at the very least, constrained to vary within a small range near the determined values.
2. There is also a high degree of peak overlap between C3S and the next most abundant phase (C2S) normally present in clinker at about 10 to 15wt%. A separate chemical extraction, using different ratios of salicylic acid and methanol, allows the removal of only C3S leaving a residue (Residue #1) in which the major phase is C2S. Collection of XRD data from this material allows refinement of the crystallite size and unit cell dimensions for C2S. This is a critical step in the development of a robust method since the high degree of overlap between the C3S and C2S patterns leads to ambiguities in the partitioning of peak intensity, and hence the derived phase abundances, between the two major silicate phases.
3. Since C3S is the major phase in clinker (typically present at about 50 to 70wt%), its parameters, including crystallite size and unit cell dimensions, can be refined by collecting XRD data from a representative sample of the plant’s clinker production. During this step, the unit cell and crystallite size parameters for the other major phases are fixed at the values determined during steps 1 and 2 above.
4. Any other materials added to clinker to make Portland cement, especially gypsum (CaSO4.2H2O) must also be defined crystallographically. Care must be taken in the sample preparation stage as gypsum can partially dehydrate to hemihydrate (CaSO4.½H2O) and anhydrite (CaSO4) during grinding. Since all three calcium sulphate phases may be present in finished cement, samples of hemihydrate and anhydrite can be prepared by heating two sub-samples of the gypsum to 125 and 600ºC respectively. The unit cell parameters and crystallite sizes of the three sulphate phases can be refined from respective XRD data sets and fixed for on-line use.
Alkali Sulphates
Since the silicate phases (C3S and C2S), present in total at up to about 85wt%, normally dominate clinker XRD patterns, all minor phases are significantly concentrated in Residue #2. This includes the important alkali sulphate phases which can (i) affect setting times and final strength, and (ii) be used to assess kiln operating conditions. Since these are normally present at a total of about 0.5wt% in clinker, and are often distributed across several Na and K sulphate phases, they are not easily identified in raw clinker patterns. However, their presence may be more easily detected in Residue #2. By optimizing the parameters of the alkali sulphates from Residue #2 data, and then constraining them in the on-line analysis system, these phases can be measured at the < 0.5wt% level (Madsen, Scarlett and Storer 2001, unpublished results) even when rapidly collected on-line data is used. In many cement plants, the major alkali sulphate phase identified is arcanite (K2SO4) although aphthitalite K3Na(SO4)2 and langbeinite K2Mg2(SO4)3) are present in cement from some plants.
The inclusion of the alkali sulphate phases in an automatic, on-line analysis method has the potential to produce meaningless results, especially under conditions where data quality may not be optimal for the analysis of minor phases. Therefore it is essential that some verification of the results be obtained. Comparison of the concentration of selected elements (for example, K2O) measured using chemical (e.g. XRF) analysis with values calculated from the quantitative phase abundances measured is a useful way to confirm the analysis.
Prior to the inclusion of the alkali sulphates in the Rietveld analytical method, there is often clear underestimation of, say, K2O calculated from the XRD results relative to the XRF. After appropriate alkali sulphate phases are included in the method, there is generally much better agreement between the calculated and observed K2O values. This agreement gives both the analyst and plant operators confidence that the phase abundances, even at this low level, are accurate and can be used to either control plant parameters or predict downstream properties of the material.
For Portland cement, the removal of the major phases to leave a residue in which the minor phases are concentrated relies on a chemical extraction process. Clearly, this approach will not be suitable for all phase systems. In some cases, concentration of minor phases can be achieved by magnetic, density or grain size separation from the major phases. Whichever method is selected, the importance of obtaining detailed parameters from all phases in the material cannot be overstated if a reliable and stable on-line analysis regime is to be achieved.
REFERENCES
Madsen I.C. and Scarlett N.V.Y. (2000) “Cement: Quantitative Phase Analysis of Portland Cement Clinker”, Chapter 16 in “Industrial Applications of X-ray Diffraction”, Marcel Dekker, ISBN 0-8247-1992-1.
Manias C., Madsen I.C. and Retallack D. (2001) “Plant Optimisation and Control Using Continuous On-line XRD for Mineral Phase Analysis”, ZKG International, 54(3),138-145.
Scarlett N.V.Y., Madsen I.C., Manias C. and Retallack D. (2001) “On-line X-ray Diffraction for Quantitative Phase Analysis: Application in the Portland Cement Industry”, Powder Diffraction, 16(2), 71-80.
Taylor H.F.W, (1990) “Cement Chemistry”, Academic Press, ISBN 0-12-683900-X
Ian C. Madsen and Nicola V.Y. Scarlett
CSIRO Process Science and Engineering
I
The emergence of interest in XRD as a tool for monitoring mineralogical abundances in industrial mineral processing plants has led to the development of a range of plant based instrumentation for the analysis of dry sample materials. These instruments can be divided into two general types, namely:-
- Those which operate in ‘batch’ mode and rely on automatic preparation and presentation of a small (5-10gm) sub-sample of the process material. Sequential XRD data is collected and QPA derived using an optimised automatic analysis regime.
- Those which take a stream of material, sub-sampled from the main product line, and pass it continuously through the instrument. Simultaneous data is collected from the moving sample and, once again, analysed using automated methods.
Portland cement is a mineralogically complex material consisting of some 10 to 15 phases of interest varying in concentration from 60 wt% down to ~ 0.2 wt%. The physical and chemical characteristics of the component phases, including the detailed chemical composition, crystallite size and ratio of polymorphs, are dependent on the nature of the starting materials and the production conditions present in a specific plant (Taylor, 1990; Madsen and Scarlett, 2000). Therefore, a robust XRD analytical regime is required to minimize the possibility of erroneous results being produced. In addition, use in an industrial processing plant often means that expert analysts are not available to check that the method is working correctly; this requires that the method operate reliably and report the results to the plant control room without operator intervention. Careful optimization of the analysis conditions and verification of results is essential.
In order to produce the most reliable analysis under these conditions, especially for the minor phases, the steps needed to develop a robust Rietveld-based analytical method include:-
1. The ferrite and aluminate phases, C4AF and C3A, are present in cement clinker at about 10 to 15 wt% in total (The cement industry nomenclature has been used when referring to the phases present in cement and clinker.
C = CaO, S = SiO2, A = Al2O3, F = Fe2O3). Hence, their diffraction patterns are normally dominated by the more abundant silicate phases, C3S and C2S. Taylor (1990) describes a method where C3S and C2S can be extracted from the sample using a mixture of salicylic acid and methanol (SAM) leaving a residue (denoted Residue #2) in which C4AF and C3A are the major phases remaining. Collection and analysis of high quality, laboratory based XRD data from this residue allows refinement of parameters for the C4AF and C3A. In addition, this pattern can be used to identify whether (i) C3A is present as more than one polymorph - cubic and orthorhombic are the most commonly occurring forms, and (ii) C4AF exhibits a range of compositions with varying Fe/Al ratios – some cement plants produce C4AF with two or more distinct compositions which have different unit cell dimensions and are readily analysed as separate phases. During on-line analysis, the crystallite size and unit cell parameters are fixed at the values determined from the residues or, at the very least, constrained to vary within a small range near the determined values.
2. There is also a high degree of peak overlap between C3S and the next most abundant phase (C2S) normally present in clinker at about 10 to 15wt%. A separate chemical extraction, using different ratios of salicylic acid and methanol, allows the removal of only C3S leaving a residue (Residue #1) in which the major phase is C2S. Collection of XRD data from this material allows refinement of the crystallite size and unit cell dimensions for C2S. This is a critical step in the development of a robust method since the high degree of overlap between the C3S and C2S patterns leads to ambiguities in the partitioning of peak intensity, and hence the derived phase abundances, between the two major silicate phases.
3. Since C3S is the major phase in clinker (typically present at about 50 to 70wt%), its parameters, including crystallite size and unit cell dimensions, can be refined by collecting XRD data from a representative sample of the plant’s clinker production. During this step, the unit cell and crystallite size parameters for the other major phases are fixed at the values determined during steps 1 and 2 above.
4. Any other materials added to clinker to make Portland cement, especially gypsum (CaSO4.2H2O) must also be defined crystallographically. Care must be taken in the sample preparation stage as gypsum can partially dehydrate to hemihydrate (CaSO4.½H2O) and anhydrite (CaSO4) during grinding. Since all three calcium sulphate phases may be present in finished cement, samples of hemihydrate and anhydrite can be prepared by heating two sub-samples of the gypsum to 125 and 600ºC respectively. The unit cell parameters and crystallite sizes of the three sulphate phases can be refined from respective XRD data sets and fixed for on-line use.
Alkali Sulphates
Since the silicate phases (C3S and C2S), present in total at up to about 85wt%, normally dominate clinker XRD patterns, all minor phases are significantly concentrated in Residue #2. This includes the important alkali sulphate phases which can (i) affect setting times and final strength, and (ii) be used to assess kiln operating conditions. Since these are normally present at a total of about 0.5wt% in clinker, and are often distributed across several Na and K sulphate phases, they are not easily identified in raw clinker patterns. However, their presence may be more easily detected in Residue #2. By optimizing the parameters of the alkali sulphates from Residue #2 data, and then constraining them in the on-line analysis system, these phases can be measured at the < 0.5wt% level (Madsen, Scarlett and Storer 2001, unpublished results) even when rapidly collected on-line data is used. In many cement plants, the major alkali sulphate phase identified is arcanite (K2SO4) although aphthitalite K3Na(SO4)2 and langbeinite K2Mg2(SO4)3) are present in cement from some plants.
The inclusion of the alkali sulphate phases in an automatic, on-line analysis method has the potential to produce meaningless results, especially under conditions where data quality may not be optimal for the analysis of minor phases. Therefore it is essential that some verification of the results be obtained. Comparison of the concentration of selected elements (for example, K2O) measured using chemical (e.g. XRF) analysis with values calculated from the quantitative phase abundances measured is a useful way to confirm the analysis.
Prior to the inclusion of the alkali sulphates in the Rietveld analytical method, there is often clear underestimation of, say, K2O calculated from the XRD results relative to the XRF. After appropriate alkali sulphate phases are included in the method, there is generally much better agreement between the calculated and observed K2O values. This agreement gives both the analyst and plant operators confidence that the phase abundances, even at this low level, are accurate and can be used to either control plant parameters or predict downstream properties of the material.
For Portland cement, the removal of the major phases to leave a residue in which the minor phases are concentrated relies on a chemical extraction process. Clearly, this approach will not be suitable for all phase systems. In some cases, concentration of minor phases can be achieved by magnetic, density or grain size separation from the major phases. Whichever method is selected, the importance of obtaining detailed parameters from all phases in the material cannot be overstated if a reliable and stable on-line analysis regime is to be achieved.
REFERENCES
Madsen I.C. and Scarlett N.V.Y. (2000) “Cement: Quantitative Phase Analysis of Portland Cement Clinker”, Chapter 16 in “Industrial Applications of X-ray Diffraction”, Marcel Dekker, ISBN 0-8247-1992-1.
Manias C., Madsen I.C. and Retallack D. (2001) “Plant Optimisation and Control Using Continuous On-line XRD for Mineral Phase Analysis”, ZKG International, 54(3),138-145.
Scarlett N.V.Y., Madsen I.C., Manias C. and Retallack D. (2001) “On-line X-ray Diffraction for Quantitative Phase Analysis: Application in the Portland Cement Industry”, Powder Diffraction, 16(2), 71-80.
Taylor H.F.W, (1990) “Cement Chemistry”, Academic Press, ISBN 0-12-683900-X
Ian C. Madsen and Nicola V.Y. Scarlett
CSIRO Process Science and Engineering
I